Visual Servoing-Based Active Vision for 3D Object Reconstruction
Ekrem Misimi, Sverre Herland, François Chaumette
AI summary
Problem
Most active vision strategies decouple viewpoint planning from control or rely on discrete global sampling and prior object knowledge, hindering efficient, real-time 3D reconstruction of unknown objects.
Approach
The authors propose a dual-task visual servoing framework that continuously couples a primary gaze-keeping task with a secondary gradient-based Next-Best-View task, driving smooth camera motion over a hemispherical surface using only local visual feedback.
Key results
- Real-time closed-loop active vision on a physical eye-in-hand robot
- NBV-object-agnostic gradient control requiring only four local evaluations
- Smooth exploration trajectories that maintain object focus and optimal depth distance
- Reconstruction coverage and efficiency highly competitive with global sampling baselines
Why it matters
Provides a computationally efficient, model-free alternative for autonomous robotic scanning, making real-time 3D reconstruction accessible without prior object data or heavy global search.
Abstract
In this letter, we present a novel dual-task, closed- loop, visual servoing-based active vision framework in an eye- in-hand configuration. The proposed active vision framework continuously drives the camera motion by coupling continuous Next-Best-View (NBV) planning and visual servo control within a unified formulation, is NBV-objective-agnostic, and enables real- time, closed-loop exploration of objects. We demonstrate how this approach can be applied to the 3D reconstruction of static volumetric objects. The approach is validated in the real world with a diverse set of relevant objects and we observe that the visual servo scheme produces smooth exploration trajectories that keep the camera focused at the object. We also show that our gradient-based continuous NBV-strategy is highly competitive with baseline strategies that leverage global viewpoint sampling and results in efficient exploration with strong object coverage.